We examined the performance of patients with probable Alzheimer's disease (AD) who participated in the placebo arm of a 12-week clinical trial. Assessments occurred at baseline and at 4, 6, 10, and 12 weeks after starting study medication. Patients received the Alzheimer's Disease Assessment Scale–Cognitive on all five occasions and the Mini-Mental State Examination on three occasions (baseline, week 6, and week 12). Their primary caregivers completed the Progressive Deterioration Scale on all five visits. This self-administered questionnaire consists of 29 items regarding the patient's behavioral and functional status. The extrapolated yearly rate of change of the mental status examinations was considerably smaller than expected compared with uncontrolled natural history studies. In addition, there was a large within-subject variability on all three measures. Scores from the baseline assessments contributed more to the variability than did the later assessments. The small magnitude of change, as well as the large between-subjects and within-subject variability, points out the methodologic drawbacks of short-duration therapeutic trials in patients with AD and the challenges in demonstrating reliable treatment effects for potential therapeutic agents.